Is the JV a poor mans' Super JX?
This came up on the JV 880 review page, so I thought it would be better here than clog up the review page with what in essence is a classic synth shoot-out - analog v digital.
I've owned and/or played most of the JV/XP (JV880/2080, XP30/50/60) line and the JX (3P/8P/10, MKS70) line.
So I decided to bench test a JV engine (2080) against a JX engine (MKS 70), mainly for fun, but I was curious, just how close can it come to the super big analog sound the Super JX is rightly famous for?
The obvious similarities are they're both Roland, they both generate sound and respond musically to performance controls, and they can both throw 4 sound sources per voice into the mix - other than that, well, one's analog (technically a digital/analog hybrid - (refer to the SCI Prophet 5 manual for that definition)) , the other is completely digital. Another similarity is they are both known for great pads. Also, they can both use MIDI velocity to mix sound sources, in which I think the Super JX is unique as an analog - to my knowledge there is no other vintage analog that can use velocity to mix or switch sound blocks.
For the test I used a JV 2080 and the SR-JV80-04, and an MKS 70 with PG 800. I initialised patches both machines, defeated internal effects, and ran them into channel strips with gates, comps and limiters only. I ran test tones to balance the inputs to minimise any psycho-accoustic effects. Testing was three fold - 1: oscillators and basic timbres (static test), 2: VCF vs TVF (static test), and 3: patch mimicking from JX to JV (dynamic test). I'll probably do some more testing, and If anyone has any ideas they want to check out, let me know.
Oscillator overview: JX uses DCOs with four wave forms; Saw, Pulse (fixed duty cycle), Square (fixed duty cycle) and white noise. The JV uses word oscillators with basic wave ROM of (880=4Meg/2080=8meg) with over 150 meg of ROM on various cards. JX DCOs can modulate each other in pairs, (hard sync, x-mod and fm). 880 has no ring mod, but 1080 and above have ring modulators, but only the JD 990 has hard sync amongst Roland digitals.
Sawtooth waves from the stock JV are essentially brighter than the JX, however, SR-JV80-04/22 (ARP Saw) is a tonal match for the JX saw. Subjectively, the JX is, well, fatter, but nothing EQ wouldn't fix. I suspect that the JX has a lower frequency response - it certainly is much noisier than the JV.
Pulse waves - this was a bit harder to match - both machines have fixed duty cycle pulses, none of which are the same, but SR-JV80-04/63 (ARP Pulse 2) is very close.
Square waves - well, I really couldn't tell any difference between either machine using the internal Synth Square wave, and the same went for noise, really couldn't pick between them.
I've been asked about sync waves, I haven't tested that yet, but there is no JV stock wave, so I'm looking at 04/131 (P5 Unisync), 04/157 (P5 X Mod) and 04/170 to 174 (MKS80 Xmod 1-5). The best hard sync waveform in the library is on SR-JV80-06/146 (Acid Sync).
So on basic synth timbres, the JV can mimic the JX pretty closely, to darn near perfectly.
Filter overview: The SuperJX has the same low pass filters as a Rev5 MKS80, so they are pretty sweet, and they are the only unmodified JX filters that will run to self-osscilation. The 16bit TVFs of the JV aremulti-modal digital filters, so quite a different technology, but are they sweet?
Starting with filters fully open, so at 100/90/75/50/25/10/0 percentiles, in terms of the timbral and dynamic changes to the sound, I could really not pick it. The changes to both tone and volume were consistent on both platforms. Next I tested filter resonance. The percentiles were reversed, so 0/10/25/50/75/90/100. From 0% to 90% resonance, the TVFs actually responded more or less exactly as the VCFs. At 100%, there were differences. I had to dial back the TVF to around 118 (out of 127) to get the same timbre as the VCF at 100%. Over 120, the TVF self oscillates like crazy, some fundamentals were generating multiple partials.
At this point I'm surprised. I shouldn't be, I mean Roland software developers probably went to great lengths to get the TVF to mimic the characteristics of a Roland VCF, but I'm still surprised. I was expecting there to be a significant difference - I mean, it's digital.....but.....it can't be better, can it?
Well, I suppose the answer is yes and no - it's a well modelled filter, it's clean, and it's accurate, it doesn't have a ton of character, but it's actually quite good. So tonally, and in theory, building a Super JX sound isn't beyond the realms.
Patch Creation: once I had the waves sorted and I was convinced the filter essentially sounded the same (even writing that I'm shaking my head) I grabbed a fairly typical Super JX patch - high stringish on B, Polybrassish on A, slow attack and release times, think Bladerunner soundtrack kinda sound. I assigned JV tones 1 & 2 to JX tone B and tones 3 & 4 to JX tone A (so read the same way). Five minutes later and I had a working facsimile, both tones used saw waves. Very little movement in the sound, but the tones were close enough, so another ten minutes or so, using both JV LFOs (so 8 in total) to introduce little variations on pitch, volume, filter and pan, very mild pitch EGs on tones 2 & 4, I had something close. Still no cigar though, so I left the bench, and returned one half hour later aftre refreshing my ears. I'd created a patch based on a JX patch, which didn't sound quite right, the stringish bit sounded more like an Alpha Juno than a JX, and the brass still needed some tweaking, but it worked as a tech demonstrator, and was actually not bad in it's own right.
So at a pinch, I'd say definitely very close to a Super JX in capability, and if not exactly the same sound, the JV has enough of the Roland character to pull it off in the mix.